Microsoft’s Private ChatGPT

There are numerous open-source LLM models out there, yet GPT-4 is still the most sought-after model. Despite being in high demand, not everyone uses it. The limitation of ChatGPT is that it’s closed-source and can’t be trusted with data and risks exposing confidential intellectual property. Data leaks in the past ascertain the “ChatGPT is untrustworthy” theory.

Microsoft, the backer of OpenAI, took notice of it and launched Azure ChatGPT, which provides similar user experience to ChatGPT but offered as the private ChatGPT. The tech giant markets itself as a platform that guarantees the privacy of data and is fully isolated from those operated by OpenAI.

This new development shows the interest of Microsoft in enterprises. They want to capture them as customers, which OpenAI failed to do because of security issues.

To deploy and operate ChatGPT on Azure, an Azure subscription is required with access to the Azure OpenAI service.?

Azure ChatGPT is built on top of a suite of technologies such as Node.js 18, Next.js 13, NextAuth.js, Langchain and many more.?

With the launch of Azure ChatGPT, Microsoft has widened the scope of its cloud service, especially Azure. In the past few years, Azure is inching closer to its competitor AWS and continuously closing the gap.?

In the recent earnings call, Microsoft CEO Satya Nadella announced that Microsoft Cloud surpassed $110 billion in annual revenue with Azure accounting for more than 50% of the total revenue, putting Azure sales at $55 billion or more.

While Microsoft is getting big on generative AI and trying to expand its cloud services, it is also encroaching upon OpenAI’s business, despite being its largest funder. Last month, Microsoft and Meta announced support for the Llama 2 family of large language models (LLMs) on Azure and Windows.?

All these developments indicate that Microsoft is exploring all the options out there and doesn't want to pull all their eggs in one basket.

Read the full story here.

Okta Sees Business in India

Every time a user logs into OpenAI's ChatGPT, Okta profits through Auth0, its cloud-based authentication and authorization platform. Okta, an Identity and Access Management firm, offers Customer Identity and Access Management (CIAM) to OpenAI, has opened its office in India. The company’s entry into India and focus on digital identity solutions align with growing cybersecurity demand.?

In an exclusive interview with AIM, Ben Goodman, Okta's Senior VP and GM for Asia Pacific and Japan, discussed their OpenAI partnership and security initiatives. Emphasizing multi-factor authentication's benefits, he highlighted reduced ransomware attacks. Okta's collaboration with OpenAI not only safeguards operations but also contributes insights that enhance security measures for other clients, reinforcing cybersecurity practices.

Read the full story here.

Snowflake Bets Big on GenAI?

Snowflake's exceptional growth journey began in 2012, amassing a diverse clientele from startups to Fortune 500 giants like Adobe, Airbnb, and Pepsico. As the age of generative AI unfolds, Snowflake aims to extend this technology to its clients.?

Primarily, generative AI enhances productivity for Snowflake's developers, similar to Microsoft's co-pilot and AWS's Whisper. Furthermore, Snowflake employs generative AI to bolster conversational search capabilities, aiding users in identifying pertinent datasets for their business needs.

While generative AI raises concerns around security and privacy, Snowflake addresses these by keeping computation within the customer's secure environment and safeguarding proprietary intellectual property within commercial LLMs. Overall, Snowflake's journey into generative AI strengthens its commitment to serving clients' evolving data-driven needs.

Read the full story here.

Pull the Plug

The online content, produced by humans, is the soul of ChatGPT, Bard, or any other generative AI platform. These platforms have been trained on the data from all over the internet, which were produced by only humans until these chatbots came into existence. These generative AI tools still need data, more and more data, but there’s a problem: they have become the victim of their own success.

Since the arrival of generative AI tools, the internet is now filled with data produced by both humans and AI. But AI needs human-produced data, because the training of AI models on AI will eventually lead to model collapse. This makes human-generated data more valuable. It’s high time humans understand their potential against machines and demand their due right from big techs which are using them as guinea pigs, or else, stop posting on the internet at all.?

Read the full story here.